Lee, W-J., and M. Mak, 1995: Dynamics of storm tracks: A linear instability perspective. Journal of the Atmospheric Sciences, 52(6), 697-723.

Abstract: The instability properties of an observed seasonal mean flow are diagnosed in terms of (i) the local versus global instability of normal modes, (ii) the convective versus absolute instability of a pulse, and (iii) the finite-time instability of optimal structures (modes), so as to learn about the dynamics of the storm tracks. The analysis is carried out in the context of the boreal winter circulation of 1982-83 with the use of a three-layer hemispheric linear balanced model.

Four clusters of normal modes with different geographical characteristics are identified among the unstable modes with an e-folding time shorter than 8 days. The Pacific modes and Atlantic modes bear a striking resemblance to the Pacific and Atlantic storm tracks of that winter, respectively. The Pacific mode primarily arises from convective instability, but recycling feedback around the hemisphere is not essential. Self-seeding of such a mode stems from an upstream convergence of ageostrophic geopotential flux over the western Pacific and may be interpreted as a manifestation of weak absolute instability. The advection of perturbation energy by the diffluent basic flow is the major process of redistributing substantial energy to the far downstream region in such a mode. A typical Atlantic mode has a much more extensive spatial structure. It is a global mode because its existence requires the process of recycling feedback around the hemisphere. Absolute instability is also found possible in the observed basic flow. It gives rise to two other local modes, one over East Asia and the other over West Africa.

There are a large number of amplifying optimal modes for an optimization time up to 10 days. The 1-day optimal mode could give rise to an intensification rate for one day exceeding that of the most unstable normal mode by sixfold. An Atlantic adjoint mode would optimally evolve to approximately an Atlantic mode in about 15 days, whereas a Pacific adjoint mode would optimally evolve to a Pacific mode in less than 10 days.